December 20, 2024

OpenAI o3 reasoning models announced

OpenAIAn AI research company — created GPT, ChatGPT, DALL-E, and the o-series reasoning models. announced o3 and o3-mini on December 20, 2024 — successors to o1 with stronger math, coding, and science reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems., offered first to safety researchers.

What it was for

OpenAIThe o3 reveal closed OpenAIAn AI research company — created GPT, ChatGPT, DALL-E, and the o-series reasoning models.'s '12 Days of OpenAIAn AI research company — created GPT, ChatGPT, DALL-E, and the o-series reasoning models.' with a direct answer to Google's GeminiGoogle's family of multimodal AI models — successors to PaLM and competitors to GPT and Claude. 2.0 FlashAdobe Flash — a browser plugin for animations and video — deprecated as the web moved to HTML5. Thinking. Sam Altman said o3-mini would ship by late January and full o3 shortly after — escalating the inferenceRunning a trained model to produce predictions — as opposed to the training phase that learns weights.-time reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. arms race that DeepSeek R1 would later join with open weights.

Companies

  • OpenAI

Why it's here

o3 signaled the next generation of OpenAIAn AI research company — created GPT, ChatGPT, DALL-E, and the o-series reasoning models. reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. models before public release.

Why it mattered

It kept frontier labs competing on test-time compute, not just parameter count.

What it solved

o1-class reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. needed a successor for harder STEM, coding, and agent benchmarks.

Media

  • OpenAI
    ImageOpenAI

    OpenAI, Public domain, via Wikimedia Commons

Related